SURROGATE MODELS FOR 3D FINITE ELEMENT CREEP ANALYSIS ACCELERATION

A gas turbine and especially its blades is at the heart of any gas and oil fired power station and thus it is critical to maximize the length of their design life as a key component for successful utilization. Blades weaknesses could affect the whole process of the energy generation, contractual penalty shall be paid in this case and company reputation can be affected. Therefore, it is important to find out the weakness of blades as quickly as possible. One of the central causes of blades' weaknesses is creep, that is the permanent deformation of material under the influence of mechanical stresses and temperature. Experience gained from real field issues show that high fidelity creep routines and FEA models must be used since the early stages of component design and by field incident investigations. Using high fidelity models in the FEA creep analysis required for turbine blade design and Lifing (Creep Capability, TMF and HCF), requires a run time of 1 - 8 days for one analysis with one setting of boundary conditions. To exploit the design space based on boundary condition uncertainties, lot of design iteration are needed. To accelerate the computation time of non-linear calculations (TMF, Creep & HCF), we investigate in this paper the integration of machine learning algorithms in lifing (stress, creep strain & displacement) prediction of an internally cooled turbine blade of a large gas turbine.